return to ICG Spaces home    ICG Risk Blog    discussions    newsletters    login    

Prediction Markets move into the forecasting mainstream


Prediction Markets move into the forecasting mainstream as a "specialists’ market [that] is smarter than the broader and more turbulence susceptible big markets," thereby "attracting significant event risk monitoring attention and that it will tend to reflect presumptions driving some of the ‘smart’ money at play in broader markets."

Prediction Markets have been of interest to me ever since DARPA's excellent, but politically tone deaf, Policy Analysis Market (PAM) was pulled down in less than a day due to misplaced Congressional criticism. (A future's contract on the assassination of Jordan's King Abdullah was one of PAM's launching examples; very pertinent but very tone deaf - typical of Poindexter, resulting in his resignation.) See: Avian Flu emerges as the new lighting rod for predictive futures markets, Dec 2005.

Speculation over Poindexter's departure spawned its own prediction market:

Ironically, the aftermath of the DARPA controversy provided a vivid illustration of the power of markets to provide information about probabilities of future events. An offshore betting exchange,, listed a new security that would pay $100 if the head of DARPA, Admiral John Poindexter, was ousted by the end of August 2003. Early trading suggested a likelihood of resignation by the end of August of 40 percent, and price fluctuations reflected ongoing news developments. Around lunchtime on July 31, reports started citing credible Pentagon insiders who claimed knowledge of an impending resignation. Within minutes of this news first surfacing (and hours before it became widely known), the price spiked to around 80. These reports left the date of Poindexter’s proposed departure uncertain, which explains the remaining risk. As August dragged on, the price slowly fell back toward 50. On August 12, Poindexter then issued a letter of resignation suggesting that he would resign on August 29. On the 12th, the market rose sharply, closing at a price of 96.

Intrade's futures contracts on the date of the first Avian Flu infection in the US struck me as an excellent bit of informed prediction that could better the received wisdom of pundits and self-styled experts. See Prediction without accountability: calling the expertise and honesty of expert predictors into question, January, 2006

I agree with a colleague's assessment of Intrade (private note):

INTRADE is functioning like a window into one of the better informed layers of a global commodity market where there are in effect tiered levels of insight into events that will drive global pricing. (Metal traders know more than screen traders, metal producers know most of all especially in getting ahead of LME inventory relevant events.) We’ve seen political events play out in crude pricing, LA government policy affect sovereign credit spreads or Chinese tax changes affect metal pricing.

Wolfers and Zitzewitz describe this form of Prediction Market as:

markets where participants trade in contracts whose payoff depends on unknown future events. Much of the enthusiasm for prediction markets derives from the efficient markets hypothesis. In a truly efficient prediction market, the market price will be the best predictor of the event, and no combination of available polls or other information can be used to improve on the market-generated forecasts. This statement does not require that all individuals in a market be rational, as long as the marginal trade in the market is motivated by rational traders. Of course, it is unlikely that prediction markets are literally efficient, but a number of successes in these markets, both with regard to public events like presidential elections and within firms, have generated substantial interest...

The basic forms of these relevant contracts will reveal the market’s expectation of a specific parameter: a probability, mean or median, respectively. But in addition, prediction markets can also be used to evaluate uncertainty about these expectations.

These probability, mean or median type contracts are differentiated:

  1. Probability: "winner-takeall" contract, the contract costs some amount $p and pays off, say, $1 if and only if a specific event occurs, like a particular candidate winning an election. The price on a winner-take-all market represents the market’s expectation of the probability that an event will occur (assuming risk neutrality)...
  2. Mean: "index" contract, the amount that the contract pays varies in a continuous way based on a number that rises or falls, like the percentage of the vote received by a candidate. The price for such a contract represents the mean value that the market assigns to the outcome...
  3. Median: "spread" betting, traders differentiate themselves by bidding on the cutoff that determines whether an event occurs, like whether a candidate receives more than a certain percentage of the popular vote.

Corporate forecasting, dodgy in all but the best of cases, should be able to benefit from internal Prediction Markets that tap the incipient organization (how work gets done) within a firm as opposed to its organizational chart (how work is theoretically done):

Spengler’s state birth stage is the organizational incipient stage, when a nucleus of highly competent/motivated individuals join to perform a task, each knowing and depending upon the other. High performance teams are incipient organizations within larger, often sclerotic, organizations. I maintain that incipients reform out of desperation, if nothing else, than to build the "Go-to" network where timely, accurate answers can be found. Incipient organizations that reside within larger networks maintain contact with, and relationship to, the larger group, but are often invisible and many of its members do not even recognize that they are seen as a key node by others. (I have seen instances where a less competent external party recognizes an incipient member and leaches off their knowledge without recompense or recognition.)

My colleague was on the same ground (private note):

[The] application of expert insight capture to Mittal’s internal price planning, [which when combined with] internal web logs seems to me to be very interesting uses of IT to radically flatten senior management’s access to real in-house expertise that is otherwise usually out of reach. It is not necessarily high value insight but it is probably better to have it than forego it. It is probably not enough to satisfy a reputation risk objective, such as a Know Your [X] due diligence requirement.

Used wisely, the combination of internal prediction markets and other tools has the potential of revealing potent social networks that could (should) become the basis for employee elevation. (See the Krebs social networking items in Globally dispersed, indigenously sited communities of terrorists upgrading to locally produced chembio agents, Nov 2006.)

I am not alone in predicting that this 'wisdom of specialized crowds' will be closely watched. With regards to its use as predictor of 2008 presidental and Congressional seats, I immediately wondered to colleagues:

how it can be spoofed so as to skew the results, thereby possibly influencing the pundits, who might then influence the electorate. If one is willing to 'lose money' on the trade, there is the possibility of gaining the 'greater good' of your candidate's victory. The first comment in the readers reply to Leonhardt's article speaks to the same idea.

Reader Responses to the Column
New York Times

Published: February 14, 2007

Odds Are, They’ll Know ’08 Winner
New York Times
February 14, 2007

Prediction Markets
Justin Wolfers and Eric Zitzewitz
Journal of Economic Perspectives, Volume 18, Number 2, Pages 107
Spring 2004

The Furor Over 'Terrorism Futures'
By Justin Wolfers and Eric Zitzewitz
Washington Post
Thursday, July 31, 2003

If This Is Harebrained, Bet on the Hare
Michael Schrage and Sam Savage
Washington Post
August 3, 2003
Original scrolled to archive

Gordon Housworth

InfoT Public  Risk Containment and Pricing Public  
In order to post a message, you must be logged in
message date / author

There are no comments available.

In order to post a message, you must be logged in